Citation: Giuliani, G.; Rodila, D.;

Külling, N.; Maggini, R.; Lehmann, A.

Downscaling Switzerland Land

Use/Land Cover Data Using Nearest

Neighbors and an Expert System.

Land 2022, 11, 615. https://doi.org/

10.3390/land11050615

Academic Editors: Víctor Hugo

González-Jaramillo and Antonio

Novelli

Received: 22 February 2022

Accepted: 20 April 2022

Published: 21 April 2022

Publisher’s Note: MDPI stays neutral

with regard to jurisdictional claims in

published maps and institutional affil-

iations.

Copyright: © 2022 by the authors.

Licensee MDPI, Basel, Switzerland.

This article is an open access article

distributed under the terms and

conditions of the Creative Commons

Attribution (CC BY) license (https://

creativecommons.org/licenses/by/

4.0/).

land

Article

Downscaling Switzerland Land Use/Land Cover Data UsingNearest Neighbors and an Expert SystemGregory Giuliani 1,2,* , Denisa Rodila 1,2, Nathan Külling 1, Ramona Maggini 3 and Anthony Lehmann 1

1 EnviroSPACE Laboratory, Institute for Environmental Sciences, University of Geneva, Bd. Carl-Vogt 66,1205 Geneva, Switzerland; denisa.rodila@unepgrid.ch (D.R.); nathan.kuelling@unige.ch (N.K.);anthony.lehmann@unige.ch (A.L.)

2 GRID-Geneva, Institute for Environmental Sciences, University of Geneva, Bd. Carl-Vogt 66,1205 Geneva, Switzerland

3 Agroscope, via A Ramél 18, 6593 Cadenazzo, Switzerland; ramona.maggini@agroscope.admin.ch* Correspondence: gregory.giuliani@unige.ch

Abstract: High spatial and thematic resolution of Land Use/Cover (LU/LC) maps are centralfor accurate watershed analyses, improved species, and habitat distribution modeling as well asecosystem services assessment, robust assessments of LU/LC changes, and calculation of indices.Downscaled LU/LC maps for Switzerland were obtained for three time periods by blending twoinputs: the Swiss topographic base map at a 1:25,000 scale and the national LU/LC statistics obtainedfrom aerial photointerpretation on a 100 m regular lattice of points. The spatial resolution of theresulting LU/LC map was improved by a factor of 16 to reach a resolution of 25 m, while thethematic resolution was increased from 29 (in the base map) to 62 land use categories. The methodcombines a simple inverse distance spatial weighting of 36 nearest neighbors’ information and anexpert system of correspondence between input base map categories and possible output LU/LCtypes. The developed algorithm, written in Python, reads and writes gridded layers of more than64 million pixels. Given the size of the analyzed area, a High-Performance Computing (HPC) clusterwas used to parallelize the data and the analysis and to obtain results more efficiently. The methodpresented in this study is a generalizable approach that can be used to downscale different types ofgeographic information.

Keywords: land cover; land use change; downscaling approach; Switzerland; geographic informationsystem; aerial photo interpretation; topographic map; inverse distance weighting; expert system

1. Introduction1.1. Pressures on Land Resources in Switzerland

The Swiss Federal Department of Environment, Transport, Energy and Communica-tions (DETEC) in its 2016 Strategy stated that by 2030, Switzerland is aiming at becominga sustainable country while remaining an attractive and competitive business locationwith a high quality of life [1]. This ambitious objective is presently challenged by severaltrends such as population growth, increased mobility, energy demand, high consumptionof resources, urbanization, loss of biodiversity and associated ecosystem services, and thedigitalization of society along with related big data [2]. These trends have an importantimpact on the environment. Therefore, protecting the environment is a central mission forthe Swiss Government, who wants to promote and adopt more sustainable approachesfor the exploitation of natural resources [3]. To this end, actions such as protecting naturalresources, improving urban planning, reducing emissions of greenhouse gases, preserv-ing water quality, retaining biodiversity and ecosystem services, protecting soils, andpreserving countryside are essential [4–6].

All these trends are also placing unprecedented demands on land. Between 1985and 2009, 15% of the country’s surface area changed [7]. Settlement and urban areas have

Land 2022, 11, 615. https://doi.org/10.3390/land11050615 https://www.mdpi.com/journal/land

Land 2022, 11, 615 2 of 21

expanded, agricultural areas have been lost, forested areas have increased, and glaciershave receded [8–10]. During the last 50 years, it is estimated that human activities haveaffected globally about 83% of the terrestrial land surface and have degraded about 60%of services provided by ecosystems. Land degradation is now at a critical point and willundermine the well-being of 3.2 billion people by 2050 [11]. Consequently, Land Cover(LC) and Land Use (LU) changes are considered as a major tangible indicator of the humanfootprint [12].

To preserve its potential to deliver goods and services, land should be efficiently andsustainably managed. National policies such as the Green Economy, the Spatial PlanningAct, the Spatial Strategy for Switzerland, the Sustainable Development Strategy 2016–2019,or the Strategy on Biodiversity are essential components to support this vision [13]. Theygenerally acknowledge that a given area of land can offer many environmental, social,cultural, and economic benefits at once. However, most ecosystems are being degraded byunsustainable exploitation, fragmentation, urban growth and development of transport,and energy networks. This reduces the spatial and functional coherence of the landscapeand consequently, degraded ecosystems are unable to provide the same services as healthyecosystems [14].

Detailed and accurate knowledge on Land Use and Land Cover Change (LU/LCC) iscrucial for many scientific and operational applications, such as watershed analyses [15,16],land use impact on stream ecology [17–19], species and habitat distribution modeling [20],dynamic modeling of species migration [21,22], reserve site selection [23], impact assess-ment on biodiversity [24], land use planning [25], or monitoring of land use changes [26].LU/LC affects many aspects of policy and decision-making processes related to climate,water, biodiversity, ecosystems, agriculture, or disasters. Additionally, LU/LCC assess-ment also contributes to many Multilateral Environmental Agreements (MEAs) and GlobalEnvironmental Goals (GEGs) to guide and assess progress toward policy outcomes [27,28].The importance of sustainable management of land resources is recognized in regionaland global policies such as the 2030 Agenda for Sustainable Development, which containsland-related targets and indicators under 14 out of the 17 Sustainable Development Goals(SDGs) [29–31]. Many land organizations and stakeholders are committed to fully imple-ment the SDGs and to monitor the land-related indicators to promote responsible landgovernance. Land is a significant resource for many sectors; timely and high-resolutionLU/LC data therefore constitute critical information for the achievement of the SDGs [32].Accurate and up-to-date LU/LCC information and related changes are also the base of asustainable development assessment [33] based on structural (both temporal and spatial)and functional (social, ecological, economical) attributes of the landscape [34]. A supple-mentary challenge is represented by the spatial scale at which the assessment is performed.For each problem under study, an appropriate scale must be identified, especially whenrelating ecological processes to landscape patterns [35].

1.2. Land Use/Land Cover Data in Switzerland

LU/LCC is increasingly acknowledged as both a driver and a consequence of climateand biodiversity changes [36–38]. This important role has been featured by the fact that landcover is considered as an Essential Climate Variable (ECV), a supplementary Essential WaterVariable (EWV), and a candidate Essential Biodiversity Variable (EBV) [39–42]. LU/LCCaffects the biophysics, biogeochemistry, and biogeography of both the atmosphere andbiosphere, with important consequences for human well-being. Consequently, accurateand timely information is necessary for understanding the impact of LU/LCC variations onthe structure and functioning of ecosystems, as well as provision, support and regulationof goods and services [29,43,44]. However, it is recognized that inadequate informationon LU/LC and its change over time is a recurrent and common problem that preventspolicymakers from making sound, informed decisions [27,45–47]. Currently, the officialLU/LC information in Switzerland (Arealstatistik) is updated approximately every 6 to8 years and derived by visual interpretation of aerial photographs where an LC and an

Land 2022, 11, 615 3 of 21

LU category are assigned to each intersection point of a regular 100 m grid [8]. Althoughthis data set is very useful thanks to its thematic richness, neither its low spatial resolutionnor the update frequency allow for providing accurate and timely information to depictand understand the dynamic of LU/LCC and the related impact across the country [48,49].Accurate LU/LC change assessment and effective LU/LCC projections require higherspatial (e.g., 30 m) and temporal (e.g., yearly) data products to build consistent timeseries [47,50].

1.3. Downscaling as a Possible Approach for High-Resolution LU/LC Data

Besides traditional remote sensing approaches, such as unsupervised, supervised,or object-based classifications [51–54], more advanced techniques include machine ordeep learning [55–58], new sensors with higher spatial and spectral resolution such asSentinel-2 [59,60], and automated procedures to reduce the time-consuming process of man-ual verification of data [61,62]; a possible alternative to generate high-resolution LU/LCCdata [63] is represented by downscaling techniques [64]. In many disciplines, downscalingis used to derive local scale maps from information available at coarser resolution. Clima-tologists refer to statistical downscaling [65] to describe this general approach that has beenwidely used not only for temperature and precipitation information [66,67] but also forwind speed [68] and air humidity [69]. In turn, downscaled climatic information is used inmany different applications, such as hydrological modeling [70–72], species distributionmodeling [73], and geological risk assessments [74]. However, downscaling of LU/LCCdata is not very common and has not been widely applied [64,75,76].

Several statistical approaches have been used for downscaling. For instance,Barodssy et al. [77] used fuzzy rule-based models to predict frequency distributions ofdaily precipitation; Bürger and Chen [78] compared regression methods to derive riverrunoff from large-scale climatic scenarios; Biau et al. [79] used geostatistical methods (krig-ing) to estimate rainfall; and Coulibaly et al. (2005) [67] investigated the use of temporalneural networks to downscale temperature and precipitation.

Downscaling is not restricted to climatic data and has been used with remote sensingdata to derive, for instance, soil moisture maps [80]. Species distribution modeling canalso be defined as a general downscaling approach that predicts species distributionsfrom point observations combined with spatially explicit environmental predictors [81].The term “downscaling” can also be used when creating a land use map from combinedinput layers at various scales. For example, Remm [82] used case-based predictions tomap the distribution of habitat classes from Landsat 7 ETM imagery, grayscale and colororthophotos, an elevation model, a digital base map, and a soil map.

Case-based algorithms are problem-solving methods that learn from experiences at alow level of generalization [83]. They can be considered as an Artificial Intelligence (AI)method that derives results from the data as directly as possible, without the formulationof an intermediate model. Machine learning (ML) specialists distinguish between lazylearning, which typically combines information during the problem-solving phase, andeager learning, which tends to derive a generalization and forget about raw observationsafter the learning phase [56,83,84]. Remm [82] argued that case-based methods represent apromising alternative for a large range of downscaling problems such as habitat mappingand the prediction of species’ potential distributions, especially with large and complexdatasets where generalization is difficult.

Based on these considerations, the aim of this paper is to present a lazy learningmethod, which could be assimilated to a case-based approach, for downscaling LU/LCinformation for Switzerland from a 100 m lattice of points to a 25 m resolution grid, takingadvantage of an existing 1:25,000 digital base map and building an expert system definingpossible correspondences between the base map and the land use categories. This methodis then applied for three different periods of time to assess land use and land cover change.

Land 2022, 11, 615 4 of 21

2. Materials and Methods

While land use and land cover maps are commonly derived from remote sensingor photointerpretation [85], traditional base maps of Switzerland have been available formore than a century and have been provided in digital format since the year 2000. LU/LCmaps derived from the classification of remotely sensed images often have a “salt andpepper” appearance that does not meet end-user demand [26]. Land use derived fromaerial photo interpretation can define many different classes of land use/cover categories,but the production is rather time-consuming. National base maps usually lack the thematicdetails that can be obtained from aerial photointerpretation but generally have an excellentgeographic precision to define landscape patches and linear features. Hereafter will bepresented the data inputs (Table 1), the downscaling methodology, and the expert system,together with its implementation and validation strategy.

Table 1. Data input sources.

Input Name Resolution Provider URL

Land Use Statistics(1992/97, 2004/09, 2013/18) Arealstatistik 100 m Swiss Federal

Statistical Office

www.bfs.admin.ch/bfs/en/home/services/geostat/swiss-federal-

statistics-geodata/land-use-cover-suitability.html (accessed on

10 December 2021).

National Base Map(2003, 2008) Vector 25 25 m swisstopo

www.swisstopo.admin.ch/en/geodata/maps/smv/smv25.html(accessed on 10 December 2021).

National Base Map(2021) swissTLM3D 25 m swisstopo

www.swisstopo.admin.ch/en/geodata/landscape/tlm3d.html(accessed on 10 December 2021).

2.1. Data Inputs2.1.1. Land Use Statistics

LU/LC data are generated by visual interpretation of aerial photographs taken froma Federal Office of Topography (swisstopo)’s aircraft flying at an altitude of 5000 m andtaking photos to regularly cover, over a 6-year period, the entire surface of Switzerland [8].LU/LC maps are obtained by visually interpreting and assigning a LU/LC category toeach point of a regular 100 m lattice laid over the Swiss territory, for a total of more than4 million points over the country.There are three official classifications: (1) the standardnomenclature NOAS04 (72 basic categories which are a combination of LC and LU, 17 and27 aggregation classes and 4 main domains); (2) the Land Cover nomenclature NOLC04(27 basic categories and 6 main domains); and (3) the Land Use nomenclature NOLU04(46 basic categories, 10 aggregation classes and 4 main domains). Three time periodsare currently available (1979/85, 1992/97, 2004/09), and the latest version has just beenfinalized (2013/18) [7].

Strictly speaking, this dataset is not a LU/LC map, because its categories are assignedto points at the intersection of a 100 m grid rather than indicating the predominant LU/LCwithin each hectare square (Figure 1). It was developed as land use statistics over relativelylarge zones rather than as an LU/LC map per se. It tends, however, to be often used asan LU/LCC map in many applications [86] and remains the most exhaustive source ofLU/LCC information for Switzerland. Even if this dataset is thematically more precisethan the classification commonly used in Europe—the Coordination of Information on theEnvironment Land Cover (CORINE Land Cover, CLC), which has 44 classes [87]—it suffersfrom a low spatial and temporal resolution. Indeed, LU/LC units are coarse with a spatialresolution of 1 hectare, and a lot of information is therefore aggregated with a large degreeof generalization. Consequently, various landscape features, qualities, particularities, andconfigurations cannot be correctly represented.

Land 2022, 11, 615 5 of 21

Land 2022, 11, x FOR PEER REVIEW 5 of 24

than the classification commonly used in Europe—the Coordination of Information on the Environment Land Cover (CORINE Land Cover, CLC), which has 44 classes [87]—it suf-fers from a low spatial and temporal resolution. Indeed, LU/LC units are coarse with a spatial resolution of 1 hectare, and a lot of information is therefore aggregated with a large degree of generalization. Consequently, various landscape features, qualities, particulari-ties, and configurations cannot be correctly represented.

Figure 1. Original LU/LC statistics with 72 classes (Arealstatistik) at a 100 m resolution from the 2013–2018 period.

2.1.2. National Base Map (Land Cover) The Swiss Federal Office of Topography (swisstopo) provides digital versions of na-

tional topographic base maps at a 1:25,000 scale as a landscape model in a vector format. The data model used until 2011 was called Vector 25 and was later replaced by the Swiss Topographic Landscape Model (TLM3D). Both include millions of natural and artificial landscape features, together with their position, shape, type and many other attributes [88]. This land cover information is defined in 29 categories (Figure 2). Data on linear fea-tures such as rivers, roads and rails can also be obtained separately. The national base maps (TLM3D) are the geographically most precise source of land cover information for the entire country with a geometric accuracy for different landscape features between 0.2 m and 3 m that is partially updated every year.

Figure 1. Original LU/LC statistics with 72 classes (Arealstatistik) at a 100 m resolution from the2013–2018 period.

2.1.2. National Base Map (Land Cover)

The Swiss Federal Office of Topography (swisstopo) provides digital versions ofnational topographic base maps at a 1:25,000 scale as a landscape model in a vector format.The data model used until 2011 was called Vector 25 and was later replaced by the SwissTopographic Landscape Model (TLM3D). Both include millions of natural and artificiallandscape features, together with their position, shape, type and many other attributes [88].This land cover information is defined in 29 categories (Figure 2). Data on linear featuressuch as rivers, roads and rails can also be obtained separately. The national base maps(TLM3D) are the geographically most precise source of land cover information for the entirecountry with a geometric accuracy for different landscape features between 0.2 m and 3 mthat is partially updated every year.

2.1.3. Resolution

For the analysis, all available datasets were rasterized either at a 25 m and/or at a100 m resolution to allow raster overlays. With a surface of 42,000 km2, Switzerland can bedescribed with approximately 4 million pixels at a 100 m resolution and with 64 millionpixels at a 25 m resolution.

2.1.4. Data Quality of Inputs

The two main data inputs of this study are of the highest possible quality from thetwo main national producers of geospatial data. The first one, land use statistics, has beendeveloped by the Swiss Federal Office of Statistics with state-of-the-art photointerpretation

Land 2022, 11, 615 6 of 21

methods, resulting in a very high quality of thematic resolution in the selection of land useclasses on each hectare point of the country. The second input, the national base maps, areproduced by the Swiss Federal Office of Topography and represent the official geographicrepresentation of the country at a 1:25,000 scale with very high spatial accuracy but a lessdeveloped thematic resolution than the first input.

Land 2022, 11, x FOR PEER REVIEW 6 of 24

Figure 2. LC information with 29 classes from the National Base Map swissTLM3D for 2021.

2.1.3. Resolution For the analysis, all available datasets were rasterized either at a 25 m and/or at a 100

m resolution to allow raster overlays. With a surface of 42,000 km2, Switzerland can be described with approximately 4 million pixels at a 100 m resolution and with 64 million pixels at a 25 m resolution.

2.1.4. Data Quality of Inputs The two main data inputs of this study are of the highest possible quality from the

two main national producers of geospatial data. The first one, land use statistics, has been developed by the Swiss Federal Office of Statistics with state-of-the-art photointerpreta-tion methods, resulting in a very high quality of thematic resolution in the selection of land use classes on each hectare point of the country. The second input, the national base maps, are produced by the Swiss Federal Office of Topography and represent the official geographic representation of the country at a 1:25,000 scale with very high spatial accu-racy but a less developed thematic resolution than the first input.

2.2. Downscaling Algorithm and Expert System 2.2.1. Downscaling Algorithm

The approach used for downscaling the existing land use information from 100 to 25 m relies on both the geographic precision of the 1:25,000 national topographic base map rasterized at 25 m and the detailed LU/LC categories obtained from the land use statistics available at 100 m. The developed algorithm uses inverse distance weighting combined with an expert system to assign reasonable land use categories at a finer scale (Figure 3).

Figure 2. LC information with 29 classes from the National Base Map swissTLM3D for 2021.

2.2. Downscaling Algorithm and Expert System2.2.1. Downscaling Algorithm

The approach used for downscaling the existing land use information from 100 to25 m relies on both the geographic precision of the 1:25,000 national topographic base maprasterized at 25 m and the detailed LU/LC categories obtained from the land use statisticsavailable at 100 m. The developed algorithm uses inverse distance weighting combinedwith an expert system to assign reasonable land use categories at a finer scale (Figure 3).

Land 2022, 11, 615 7 of 21Land 2022, 11, x FOR PEER REVIEW 7 of 24

Figure 3. Downscaling land use statistics (2013–2018) in the area of Zermatt from 100 m to 25 m resolution using inverse distance weighting, expert knowledge, national base map, transport and river information at 25 m. (a) Expert system (details in Figure 4), (b) 1:25,000 base map, (c,c’) hectare land use information, (d,d’) downscaled land use, (e) 1:25,000 linear features (rivers, roads and rails), (f,f’) overlay of linear features to downscaled land use at 25 m resolution. Downscaling process (p1) and linear features addition (p2).

Figure 3. Downscaling land use statistics (2013–2018) in the area of Zermatt from 100 m to 25 mresolution using inverse distance weighting, expert knowledge, national base map, transport andriver information at 25 m. (a) Expert system (details in Figure 4), (b) 1:25,000 base map, (c,c′) hectareland use information, (d,d′) downscaled land use, (e) 1:25,000 linear features (rivers, roads and rails),(f,f′) overlay of linear features to downscaled land use at 25 m resolution. Downscaling process (p1)and linear features addition (p2).

Land 2022, 11, x FOR PEER REVIEW 8 of 24

Figure 4. Expert system showing the possible authorized Landuse100 categories within BaseMap25 units; (1) represents possible choices, (2) unique choice, and (3) the default choice in case of lack of decision. Ten land use categories remain unmatched (shaded). Three categories correspond to punc-tual or linear features (dark grey), and seven categories (light grey) correspond to different types of buildings that were merged with their surroundings (green). A full version of the expert table is available in the Supplementary Material Table S1.

The main steps of the downscaling methodology are the following (data preparation: 1–2–3; process [1] of downscaling: 4–10; process [2] for linear features: 11): 1. Rasterize a land use grid at 100 m resolution from the lattice of points of the land use

statistics (Landuse100). 2. Convert to Non-Applicable (NA) land use categories that correspond to linear fea-

tures (rivers, roads, rails). 3. Rasterize the primary surfaces of the land cover vector base map at a 25 m resolution

(BaseMap25). 4. Visit each BaseMap25 pixel (target pixel). 5. Then, according to the expert system table (Figure 4), select the land use categories

that could be eligible for the target pixel. In some rare cases, assign the only possible category (then go to point 10).

6. Select among the 36 nearest Landuse100 neighbors those with eligible categories. 7. Calculate the inverse distance to each neighbor. 8. Sum up the inverse distances for each category. 9. Assign to the BaseMap25 pixel the category obtaining the higher score or, in case of

lack of decision, assign the best replacement choice according to the expert system table.

10. Repeat steps 4 to 9 for each BaseMap25 pixel. 11. Replace categories wherever river, road, or rail linear features are available from

BaseMap25. Only main roads (>3 m wide) and main railways were considered with-out tunnels and bridges. Underground rivers were ignored. Rivers, railways, roads,

Figure 4. Expert system showing the possible authorized Landuse100 categories within BaseMap25units; (1) represents possible choices, (2) unique choice, and (3) the default choice in case of lack of

Land 2022, 11, 615 8 of 21

decision. Ten land use categories remain unmatched (shaded). Three categories correspond topunctual or linear features (dark grey), and seven categories (light grey) correspond to different typesof buildings that were merged with their surroundings (green). A full version of the expert table isavailable in the Supplementary Material Table S1.

The main steps of the downscaling methodology are the following (data preparation:1–2–3; process [1] of downscaling: 4–10; process [2] for linear features: 11):

1. Rasterize a land use grid at 100 m resolution from the lattice of points of the land usestatistics (Landuse100).

2. Convert to Non-Applicable (NA) land use categories that correspond to linear features(rivers, roads, rails).

3. Rasterize the primary surfaces of the land cover vector base map at a 25 m resolution(BaseMap25).

4. Visit each BaseMap25 pixel (target pixel).5. Then, according to the expert system table (Figure 4), select the land use categories

that could be eligible for the target pixel. In some rare cases, assign the only possiblecategory (then go to point 10).

6. Select among the 36 nearest Landuse100 neighbors those with eligible categories.7. Calculate the inverse distance to each neighbor.8. Sum up the inverse distances for each category.9. Assign to the BaseMap25 pixel the category obtaining the higher score or, in case of lack

of decision, assign the best replacement choice according to the expert system table.10. Repeat steps 4 to 9 for each BaseMap25 pixel.11. Replace categories wherever river, road, or rail linear features are available from

BaseMap25. Only main roads (>3 m wide) and main railways were consideredwithout tunnels and bridges. Underground rivers were ignored. Rivers, railways,roads, and freeways were rasterized at 25 m and added in this order after the firstphase of downscaling.

The Inverse Distance Weighting (IDW) calculates a scaled distance to each of the36 Landuse100 neighbors around the pixel under investigation, and does so in two spatialdimensions (Equation (1)):

dj(i) = sqrt((x25i − x100j)2 + (y25i − y100j)2)/maxrange (1)

where j spans from 1 to 36 nearest neighbors and i represents each visited pixel at a 25 mresolution and maxrange the maximum distance between these pixels.

Then the inverse distances are summed up by land use category (Equation (2)):

Dk(i) = ∑36j=1

(λj(k)

dj(i) + s

)(2)

where λj(k) ={

1 i f landusetypej = k0 otherwise

, and s is smoothing factor.

Finally, the category scoring the highest sum of inverse distances is assigned to thepixel under investigation (Equation (3)):

K(i) = greatestk(Dk(i)) (3)

IDW is generally used for interpolating cardinal discrete values (e.g., temperatures,precipitations) but can be also used on nominal values to spatially interpolate missingLC data [89], create super-resolution LC maps [90], or rescale LC data [91]. IDW assumesthat values that are close to one another are more similar than those that are at a greaterdistance. In the proposed method, measured values surrounding the prediction locationare considered but then they are spatially weighted (i.e., taking into account the distanceof each pixel and the frequency of classes) and constrained (i.e., by the expert system).

Land 2022, 11, 615 9 of 21

Consequently, the proposed downscaling method is filling a geographically very precisemap (the base map) with information on the most plausible LU/LC category found in theneighborhood by reducing the possible choices with an expert table. This approach has theadvantage of fully respecting the geographical quality of the base map while enriching itwith a better thematic resolution.

2.2.2. Expert System

An expert system was used to constrain the possible choices when using informationfrom BaseMap25 to select the most appropriate Landuse100 category (Figure 4). Forinstance, if a forest patch is defined for a particular site in BaseMap25, the expert systemconstrains the choice of Landuse100 categories to include only those related to forest.Furthermore, the expert system can also select a default Landuse100 category when thedistance-based algorithm fails to make a clear selection because of the lack of eligible landuse categories within the searching neighborhood.

2.3. Implementation

Although the algorithm used is relatively simple, it requires extensive calculations onseveral large grids (25 m resolution grid for Switzerland contains approximately 64 millionpixels), which is time-consuming. To ensure the portability of the code on different plat-forms, the algorithm has been implemented using a set of open-source software andlibraries. The programming language is Python 3 [92] using the PyCharm CommunityEdition (CE) Integrated Development Environment (IDE) [93]. The implemented algorithmrelies on the following libraries: (1) the Geospatial Data Abstraction Library (GDAL) [94]to handle raster data; (2) Numpy [95] and math for mathematical operations on largemulti-dimensional arrays; and (3) xlrd [96] and pandas [97] for reading and formattinginformation from Excel files.

To ensure a fast and efficient processing, the algorithm has been parallelized andexecuted on the High-Performance Computing cluster at University of Geneva [98], allow-ing to process the entire 64 million pixels grid significantly faster. The analyzed area issubdivided in tiles and the processing of each tile is performed on a different node on thecluster. The surface of Switzerland was divided in 900 tiles (30 × 30), which were associatedto 900 jobs in the cluster, using SLURM job arrays. The processing time is on average lessthan 2 h per tile, but the actual processing time of a single tile is strongly dependent onthe user priority and on the cluster load at the execution time. When the results of all thejobs are available, a final script is launched to assemble the 900 processed tiles together in asingle image.

The code is freely available on GitHub [99] under an Apache 2.0 license, and a staticversion can be downloaded from the University of Geneva digital repository [100] under aCC-BY 4.0 license.

2.4. Validation and Accuracy Assessment

The categories of the downscaled LU/LC maps at 25 m (with or without linear features)for the three periods were compared with the categories of the original LU/LC pointstatistics. A random sample of 500,000 out of 4,129,070 points was used for this assessment.Multi-class classification metrics were used to assess the efficiency of the algorithm toclassify each LU/LC category [101–103]. Results are shown as F1-score, which is obtainedby harmonic mean from the recall and precision, and Cohen’s kappa coefficient [104].Results were expressed as class F1-scores and weighted means for each dataset, and kappacoefficient for each dataset. Barplots and boxplots are used to display these metrics andthe main misclassifications per category. The relative surface area per category was alsocompared to that of the original Landuse100 dataset [64,105].

Land 2022, 11, 615 10 of 21

3. Results

Results from the land use downscaling are best appreciated at a regional scale(Figures 5 and S1), where the map produced by the algorithm not only preserves the geo-graphic precision of the original BaseMap25 map but also inherits the higher definition ofcategories (66 categories in the downscaled map compared to 29 categories in BaseMap25).Seventy-two categories were originally found in Landuse100, but a few groupings had tobe performed on some less important categories (e.g., buildings and their surroundings).The new map has a much finer grain and can be used for Geographical Information System(GIS) overlays at much finer scales, where the 16-fold increase in the density of pixels givessubstantially improved results. Fine-scale details on roads and river networks were alsomaintained, whereas they are often difficult to represent adequately in coarse resolutionraster formats (Figure 6).

Land 2022, 11, x FOR PEER REVIEW 11 of 24

Figure 5. Land use downscaling at 25 m of Switzerland for the 2013–2018 period. Swiss regional parks boundaries in red. (See Figure 6 for legend interpretation.)

Figure 5. Land use downscaling at 25 m of Switzerland for the 2013–2018 period. Swiss regionalparks boundaries in red. (See Figure 6 for legend interpretation.)

The average percentage of similarity for the three periods evaluated between theoriginal points found in the Landuse100 statistics and the resulting Landuse25 classifi-cations is 71% when assessed without roads, rivers, and rails, and 68% when assessedwith these linear features. Several categories have a correspondence of 70% or greater.Overall F1-scores and kappa coefficient are high (0.69 and 0.67, respectively). Individualclasses F1-scores range from 0.007 to 0.98 depending on the category (Figure 7). Weightedmean F1-scores per dataset range from 0.67 to 0.70. Kappa scores range from 0.65 to 0.69.Categories from the Landuse100 classification that show zero correspondence are those thatwere not predicted, such as flood protection structures (63), field fruit trees (38), or groovesand hedges (58), and were thus discarded from the validation analysis. The categories thatare best respected by the downscaling are those that have a large spatial representation andthose that were imposed from the topographic map.

Land 2022, 11, 615 11 of 21Land 2022, 11, x FOR PEER REVIEW 12 of 24

Figure 6. Land use downscaling from 100 m to 25 m and overlay of linear features for state of Geneva for the three periods.

The average percentage of similarity for the three periods evaluated between the original points found in the Landuse100 statistics and the resulting Landuse25 classifica-tions is 71% when assessed without roads, rivers, and rails, and 68% when assessed with these linear features. Several categories have a correspondence of 70% or greater. Overall F1-scores and kappa coefficient are high (0.69 and 0.67, respectively). Individual classes F1-scores range from 0.007 to 0.98 depending on the category (Figure 7). Weighted mean F1-scores per dataset range from 0.67 to 0.70. Kappa scores range from 0.65 to 0.69. Cate-gories from the Landuse100 classification that show zero correspondence are those that were not predicted, such as flood protection structures (63), field fruit trees (38), or grooves and hedges (58), and were thus discarded from the validation analysis. The cate-gories that are best respected by the downscaling are those that have a large spatial rep-resentation and those that were imposed from the topographic map.

Figure 6. Land use downscaling from 100 m to 25 m and overlay of linear features for state of Genevafor the three periods.

Land 2022, 11, x FOR PEER REVIEW 13 of 24

Figure 7. Weighted boxplots showing the F1-scores from the 72 downscaled classes (Landuse25) compared to the original classes (Landuse100), for each period, with (“all”) and without linear fea-tures. Numbers along the boxplots represent the land use categories and their size the proportion of each category in the dataset. Blue represents categories attributed based on “swisstopo” original classes; black represents categories attributed based on IDW of the 36 nearest neighbors. Values under each boxplot are overall weighted mean F1-score (F1) and kappa coefficient (k) for the con-sidered year and dataset.

The percentages of misclassification for the downscaling of 2018 without linear fea-tures are represented in Figure 8. The major misclassifications concern unproductive grasslands (15%), farm pastures (12%), and meadows (10%). The classes in which the pix-els were misattributed are also presented (color in the box). This figure will help in under-standing the behavior of the downscaling algorithm to further improve it.

Figure 7. Weighted boxplots showing the F1-scores from the 72 downscaled classes (Landuse25)compared to the original classes (Landuse100), for each period, with (“all”) and without linear

Land 2022, 11, 615 12 of 21

features. Numbers along the boxplots represent the land use categories and their size the proportionof each category in the dataset. Blue represents categories attributed based on “swisstopo” originalclasses; black represents categories attributed based on IDW of the 36 nearest neighbors. Values undereach boxplot are overall weighted mean F1-score (F1) and kappa coefficient (k) for the consideredyear and dataset.

The percentages of misclassification for the downscaling of 2018 without linear featuresare represented in Figure 8. The major misclassifications concern unproductive grasslands(15%), farm pastures (12%), and meadows (10%). The classes in which the pixels weremisattributed are also presented (color in the box). This figure will help in understandingthe behavior of the downscaling algorithm to further improve it.

Land 2022, 11, x FOR PEER REVIEW 14 of 24

Figure 8. Percentage of all misclassifications for the 17 main classes of the 2018 downscaled LU without linear features. Legend lists categories that were misattributed.

The absolute surface area difference from the downscaled map (Landuse25) catego-ries compared to the original Landuse100 map ranges from −87,585 ha for the category “unproductive grass and shrubs” to +118,174 ha for the category “normal dense forest” (Figure 9). On average, a difference of 532 ha per category is observed. Underlaying rea-sons for large discrepancies in surface areas (misattributed categories) are further dis-played in Figure 8. The relative surface area difference (proportion from downscaled map compared to original) ranges from −100% for small categories that were not included in the algorithm to +355% for cat. 14, “surroundings of unspecified buildings”.

Figure 8. Percentage of all misclassifications for the 17 main classes of the 2018 downscaled LUwithout linear features. Legend lists categories that were misattributed.

The absolute surface area difference from the downscaled map (Landuse25) categoriescompared to the original Landuse100 map ranges from −87,585 ha for the category “unpro-ductive grass and shrubs” to +118,174 ha for the category “normal dense forest” (Figure 9).On average, a difference of 532 ha per category is observed. Underlaying reasons forlarge discrepancies in surface areas (misattributed categories) are further displayed inFigure 8. The relative surface area difference (proportion from downscaled map comparedto original) ranges from −100% for small categories that were not included in the algorithmto +355% for cat. 14, “surroundings of unspecified buildings”.

Land 2022, 11, 615 13 of 21Land 2022, 11, x FOR PEER REVIEW 15 of 24

Figure 9. Absolute surface area difference from the downscaled map (Landuse25) categories com-pared to the original Landuse100 map.

Figure 9. Absolute surface area difference from the downscaled map (Landuse25) categories com-pared to the original Landuse100 map.

4. Discussion

The method presented in this study is a generalizable approach that can be used todownscale different types of geographic information such as results from photo interpreta-tions, classifications of remotely sensed images, or vegetation and soil maps. The generalidea is to use the nearest precise point observations to define the attribute of a target pixelat a finer resolution within an area defined by a high-resolution land cover map. Other

Land 2022, 11, 615 14 of 21

co-variables such as remote sensing indices (e.g., Normalized Difference Vegetation Index(NDVI)), topographic position, slope, or orientation could also be used to help model themost probable LU/LC category and could certainly further improve the present method.

In the case study presented here, the use of a national topographic base map at a1:25,000 scale to define main land cover patches guarantees a perfect overlay with officialmaps. Indeed, topographic base maps are now available in digital format and are widelyused as reference maps in most studies and field work. The possibility of matching thegeometry defined by base maps with detailed land use categories reinforces the chance ofuptake from end-users. However, some recent changes in LU/LC that might be visiblefrom remote sensing images could be lost if involving changes between incompatible landuse categories as defined by the expert system.

One of the main strengths of the proposed approach is that it avoids the “salt andpepper” effect generally resulting from the classification of remotely sensed images [26],except in the case of object-oriented classifications [106]. We believe that our approachcould be particularly useful in this respect and could be used to smooth out land usemaps obtained from remote sensing classifications. In such a case, land use statisticswould be replaced by the land use classification obtained from supervised or unsupervisedclassification of the remotely sensed image, whereas the land cover information would stillbe retrieved from the national base map. Moreover, with the increased spatial resolutionand the removal of the “salt and pepper” effect, LU/LC changes are more evident and canbe assessed more easily (Figure 10).

Land 2022, 11, x FOR PEER REVIEW 17 of 24

Figure 10. Comparison of the 1997 and 2018 LU/LC data with original 100 m resolution (above) and the downscaled outputs at 25 m (below) for the Bulle and Freiburg cities. Urban expansion (as well as other land cover changes and the smoothing effect) of this economically active region can be more easily visualized with the improved spatial resolution.

The proportion (71%) of exact matches between the land use categories recorded at the 100 m resolution lattice points of the Landuse100 statistics and the corresponding points in the downscaled map (25 m resolution) is particularly encouraging. Discrepancies appear to result mainly from the change in scale and geometric precision brought in by the 1:25,000 base maps. The fundamental difference in approach between the land use statistics (punc-tual) and the topographic map (surface) is also important. A visual check confirms that most divergences mainly occur along boundaries of patches of different land use and along linear and small features. By using an inverse distance calculation—which gives a higher weight to close-by information—to choose the best land use category for each pixel, we greatly fa-vored the retaining of input land use categories at an observed location in the result. This effect can be modulated by the smoothing factor in Equation (2).

The approach could be used at even finer scales by rasterizing for instance the topo-graphic maps at 10 m instead of 25 m or using a regional map at a finer scale (1:10,000). Getting accurate land use information is crucial for calculating landscape indices such as those obtained from FRAGSTAT [107]. As demonstrated by Uuemaa et al. [108], changing grain size can have a significant effect on many landscape metrics. Therefore, one should pay attention to the scale at which a given landscape metric is calculated, and what grain

Figure 10. Comparison of the 1997 and 2018 LU/LC data with original 100 m resolution (above) andthe downscaled outputs at 25 m (below) for the Bulle and Freiburg cities. Urban expansion (as wellas other land cover changes and the smoothing effect) of this economically active region can be moreeasily visualized with the improved spatial resolution.

Land 2022, 11, 615 15 of 21

The proportion (71%) of exact matches between the land use categories recorded atthe 100 m resolution lattice points of the Landuse100 statistics and the correspondingpoints in the downscaled map (25 m resolution) is particularly encouraging. Discrepanciesappear to result mainly from the change in scale and geometric precision brought in by the1:25,000 base maps. The fundamental difference in approach between the land use statistics(punctual) and the topographic map (surface) is also important. A visual check confirmsthat most divergences mainly occur along boundaries of patches of different land use andalong linear and small features. By using an inverse distance calculation—which gives ahigher weight to close-by information—to choose the best land use category for each pixel,we greatly favored the retaining of input land use categories at an observed location in theresult. This effect can be modulated by the smoothing factor in Equation (2).

The approach could be used at even finer scales by rasterizing for instance the topo-graphic maps at 10 m instead of 25 m or using a regional map at a finer scale (1:10,000).Getting accurate land use information is crucial for calculating landscape indices such asthose obtained from FRAGSTAT [107]. As demonstrated by Uuemaa et al. [108], changinggrain size can have a significant effect on many landscape metrics. Therefore, one shouldpay attention to the scale at which a given landscape metric is calculated, and what grainsize of land use maps is best adapted for the purpose, as there is no single land use scalewhich is “best” for observing and managing changes in land use patterns.

The implementation of a case-based reasoning algorithm in Python code proved tobe a very efficient approach given the very large number of pixels (64 million) to handlebut required the development of a purpose-written software. The availability of toolsfor implementing case-based approaches within existing GIS packages would certainlybe very useful for many applications. One such tool has been developed by Remm [82]using Microsoft Visual Studio.NET, and this can predict several response types (binomial,multinomial, continuous, and complex) based on continuous and categorical predictors.

Riitters [109] mentions the practical trade-off that exists between the generality, preci-sion, and realism of a method, as previously proposed by Levins [110], and emphasizesthat generality and realism should be maximized at national scales, while precision shouldbe optimized at local scales. Indeed, the use of the same raw data to develop models atdifferent scales will enhance thematic and geographic comparisons between disciplinesand across scales. Methods to upscale and downscale data are therefore central to local,national, and global environmental assessments.

The presented approach contributes to tackle the need of increased spatial resolutionof LU/LCC information. However, the temporal issue remains a major problem. Indeed,land cover changes are caused by either natural or anthropogenic sources such as climatechange, demographic growth, and economic growth [111,112]. Therefore, the state ofland cover is highly dynamic and involves an inherent challenge for its mapping andmonitoring that remains not adequately addressed [85]. Timely and reliable informationon land cover change is crucial to efficiently mitigate the negative impact of environmentalchanges [113,114]. Traditional environmental data collection (e.g., field data collection)suffers from many shortcomings, among them the data inconsistency caused by changes inreporting methodologies through time, the gaps (missing measurements) in data series,and the fact that it is notoriously time-consuming and labor-intensive [115]. One of themain advantages of remotely sensed data collection is that it can provide a synoptic andrepetitive view of a given area/region. With the opening of different Earth Observations(EO) data archives such as Landsat [112,116,117], it becomes possible to build consistenttime series (i.e., image of the same location at regular intervals) to compare different periodsof time and derive trends [118–120].

At the national scale, various attempts have been made to use satellite imagery, butunfortunately, no processes are currently in place to routinely generate accurate, consistent,and regular LU/LCC data. These large volumes of freely and openly available EO dataare still underutilized and are not effectively used for national environmental monitoring.Therefore, mapping and monitoring LU/LC changes remain as challenges that are not

Land 2022, 11, 615 16 of 21

adequately addressed at the national scale, and new methodologies are required to produceconsistent and reliable yearly, medium-to-high-resolution (spatial, temporal, thematic) timeseries of LU/LCC data and projections of their (future) change across Switzerland to informnational and regional environmental policies and planning.

With the opening of different medium-to-high-resolution satellite EO data archivessuch as Landsat or Copernicus and the development of advanced data science techniques(e.g., Big Data, Artificial Intelligence, High-Performance Computing), it now becomespossible to build consistent time series of LC, to investigate the spatio-temporal dynamicsof LC, and to perform quantitative assessments of LC dynamics, by comparing differentperiods of time, deriving trends, and determining environmental trajectories [121]. Thiscan generate a consistent and reliable LU/LC time series that will help to understand theevolution of LU/LCC in Switzerland over the last 30 years and to model future changesaccording to plausible scenarios by 2050 [122,123].

To reach this objective, an essential pre-condition to support user applications andgenerate usable information products is to facilitate data access, preparation, and analysis.The systematic and regular provision of Analysis-Ready Data (ARD) can significantlyreduce the burden of EO data usage. To be considered as ARD, data should be processedaccording to a minimum set of requirements (e.g., radiometric and geometric calibration;atmospheric correction; metadata description) and organized in a way that allows im-mediate analysis without additional effort [124,125]. In Switzerland, more than 37 yearsof satellite EO Analysis-Ready Data over Switzerland are made available by the SwissData Cube [126,127]. The increasing availability of EO data together with the improvedcomputing and storage capacities allow monitoring, mapping, and assessing LU/LC andits change over time on large areas in a consistent and reliable manner. This favors thedevelopment of annual, high-quality LU/LC products based on time series data and caninform on class stability and transitions [128].

5. Conclusions

The proposed downscaling approach allowed for combining the geographic precisionof existing topographic base maps with the thematic details of photo-interpreted land usestatistics. Improved land use maps open the door to more accurate watershed analyses,species and habitat distribution modeling, and species dynamic models of migration.Accurate land use information is also the base for developing sustainable developmentindicators defining the structural and functional attributes of landscapes.

The proposed approach could be implemented for three time periods. It allowed for anefficiently downscaled LU/LC at a spatial resolution that is more suitable for environmentalchange monitoring. The increased spatial resolution removed the “salt and pepper” effect,and consequently, LU/LC changes are more evident. However, the changing definition ofthe base map input across the years resulted in discrepancies in the resulting downscaledmaps that are refraining their use for land use change analyses. We therefore recommendusing the original land use statistics data for trend analyses and our downscaled data forGIS analyses at each time period. The presented approach contributes to tackling the needfor increased spatial resolution of LU/LC information, but the temporal issue is still a majorproblem. The use of dense time series of satellite data such as those provided in the SwissData Cube can be a promising solution to investigate to obtain high spatial and temporalLU/LC data over the country.

Supplementary Materials: The following are available online at https://www.mdpi.com/article/10.3390/land11050615/s1. Figure S1. Final result of the downscaled Land Use/Land Cover at 25 mfor the period 2013/18. Table S1. Full version of the expert table.

Land 2022, 11, 615 17 of 21

Author Contributions: Conceptualization, A.L. and R.M.; methodology, A.L.; software, A.L., G.G.and D.R.; validation, A.L. and N.K.; formal analysis, A.L., G.G. and D.R.; data curation, A.L. andN.K.; writing—original draft preparation, G.G., A.L. and R.M.; writing—review and editing, all;visualization, A.L. and N.K.; supervision, A.L.; project administration, A.L.; funding acquisition, A.L.All authors have read and agreed to the published version of the manuscript.

Funding: Support by the Swiss Federal Office for the Environment with the ValPar.ch project isgratefully acknowledged.

Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.

Data Availability Statement: The results presented in this study are openly available in Yareta,the University of Geneva Digital Repository, at: https://doi.org/10.26037/yareta:dlx3hu54jfa3ne3c2xjfcnqpxm (accessed on 21 February 2022) and in the ValPar project Spatial Data Infrastructureat: http://valpar.unige.ch:8080/geonetwork/srv/eng/catalog.search#/metadata/da78fa9f-a756-4538-a28a-8be7d53d4676 (accessed on 21 February 2022). The source code of the algorithm is availableat: https://github.com/ggiuliani/LULCdown (accessed on 21 February 2022). Publicly availabledatasets were analyzed in this study. These data can be found here: Land Use Statistics (https://www.bfs.admin.ch/bfs/de/home/statistiken/raum-umwelt/erhebungen/area.html, accessedon 21 February 2022); Base Maps (SwissTLM3D—https://www.swisstopo.admin.ch/en/geodata/landscape/tlm3d.html, accessed on 21 February 2022).

Acknowledgments: Swisstopo and the Swiss Federal Office for Statistics are gratefully acknowledgedfor making the input data freely available.

Conflicts of Interest: The authors declare no conflict of interest.

References1. DETEC. Stratégie 2016 du DETEC; DETEC: Bern, Switzerland, 2016.2. Maxwell, S.L.; Fuller, R.A.; Brooks, T.M.; Watson, J.E.M. Biodiversity: The Ravages of Guns, Nets and Bulldozers. Nat. News 2016,

536, 143. [CrossRef] [PubMed]3. Conféderation Suisse. Swiss Position on a Framework for Sustainable Development Post-2015; BAFU: Bern, Switzerland, 2016.4. Lehmann, A.; Guigoz, Y.; Ray, N.; Mancosu, E.; Abbaspour, K.C.; Freund, E.R.; Allenbach, K.; Bono, A.D.; Fasel, M.; Gago-Silva,

A.; et al. A Web Platform for Landuse, Climate, Demography, Hydrology and Beach Erosion in the Black Sea Catchment. Sci.Data 2017, 4, sdata201787. [CrossRef] [PubMed]

5. Artmann, M.; Bastian, O.; Grunewald, K. Using the Concepts of Green Infrastructure and Ecosystem Services to Specify Leitbilderfor Compact and Green Cities—The Example of the Landscape Plan of Dresden (Germany). Sustainability 2017, 9, 198. [CrossRef]

6. Rounsevell, M.D.A.; Reginster, I.; Araújo, M.B.; Carter, T.R.; Dendoncker, N.; Ewert, F.; House, J.I.; Kankaanpää, S.; Leemans, R.;Metzger, M.J.; et al. A Coherent Set of Future Land Use Change Scenarios for Europe. Agric. Ecosyst. Environ. 2006, 114, 57–68.[CrossRef]

7. Swiss Federal Statistical Office. Land Use in Switzerland—Results of the Swiss Land Use Statistics; SFO: Neuchâtel, Switzerland, 2013.8. Swiss Federal Statistical Office. The Changing Face of Land Use: Land Use Statistics of Switzerland; SFO: Neuchâtel, Switzerland,

2001; p. 32.9. European Environment Agency. Land Cover 2012—Country Fact Sheet; EEA: Copenhagen, Denmark, 2017; p. 18.10. Office Fédéral de la Statistique. Statistique de la Superficie 2013/18; Office Fédéral de la Statistique: Neuchâtel, Switzerland, 2020.11. IPBES. Summary for Policymakers of the Thematic Assessment of Land Degradation and Restoration; IPBES Secretariat: Bonn,

Germany, 2018.12. United Nations Department of Economic and Social Affairs. Sustainable Land Use for the 21st Century; United Nations Department

of Economic and Social Affairs: New York, NY, USA, 2012; p. 82.13. State of the Environment. Environment Switzerland 2015; Swiss Federal Council: Bern, Switzerland, 2015; p. 144.14. Burkhard, B.; Kroll, F.; Müller, F.; Windhorst, W. Landscapes‘ Capacities to Provide Ecosystem Services—A Concept for Land-

Cover Based Assessments. Landsc. Online 2009, 15, 1–22. [CrossRef]15. Mander, Ü.; Kull, A.; Kuusemets, V. Nutrient Flows and Land Use Change in a Rural Catchment: A Modelling Approach. Landsc.

Ecol. 2000, 15, 187–199. [CrossRef]16. Hörmann, G.; Horn, A.; Fohrer, N. The Evaluation of Land-Use Options in Mesoscale Catchments: Prospects and Limitations of

Eco-Hydrological Models. Ecol. Model. 2005, 187, 3–14. [CrossRef]17. Snyder, C.D.; Young, J.A.; Villella, R.; Lemarié, D.P. Influences of Upland and Riparian Land Use Patterns on Stream Biotic

Integrity. Landsc. Ecol. 2003, 18, 647–664. [CrossRef]

Land 2022, 11, 615 18 of 21

18. Zhou, T.; Wu, J.; Peng, S. Assessing the Effects of Landscape Pattern on River Water Quality at Multiple Scales: A Case Study ofthe Dongjiang River Watershed, China. Ecol. Indic. 2012, 23, 166–175. [CrossRef]

19. dos Reis Oliveira, P.C.; van der Geest, H.G.; Kraak, M.H.S.; Verdonschot, P.F.M. Land Use Affects Lowland Stream Ecosystemsthrough Dissolved Oxygen Regimes. Sci. Rep. 2019, 9, 19685. [CrossRef]

20. Oja, T.; Alamets, K.; Pärnamets, H. Modelling Bird Habitat Suitability Based on Landscape Parameters at Different Scales. Ecol.Indic. 2005, 5, 314–321. [CrossRef]

21. Boone, R.B.; Hunter, M.L. Using Diffusion Models to Simulate the Effects of Land Use on Grizzly Bear Dispersal in the RockyMountains. Landsc. Ecol. 1996, 11, 51–64. [CrossRef]

22. Akçakaya, H.R. Linking Population-Level Risk Assessment with Landscape and Habitat Models. Sci. Total Environ. 2001, 274,283–291. [CrossRef]

23. van Langevelde, F.; Schotman, A.; Claassen, F.; Sparenburg, G. Competing Land Use in the Reserve Site Selection Problem. Landsc.Ecol. 2000, 15, 243–256. [CrossRef]

24. Crist, P.J.; Kohley, T.W.; Oakleaf, J. Assessing Land-Use Impacts on Biodiversity Using an Expert Systems Tool. Landsc. Ecol. 2000,15, 47–62. [CrossRef]

25. Theobald, D.M.; Hobbs, N.T.; Bearly, T.; Zack, J.A.; Shenk, T.; Riebsame, W.E. Incorporating Biological Information in LocalLand-Use Decision Making: Designing a System for Conservation Planning. Landsc. Ecol. 2000, 15, 35–45. [CrossRef]

26. Bock, M.; Rossner, G.; Wissen, M.; Remm, K.; Langanke, T.; Lang, S.; Klug, H.; Blaschke, T.; Vršcaj, B. Spatial Indicators for NatureConservation from European to Local Scale. Ecol. Indic. 2005, 5, 322–338. [CrossRef]

27. Szantoi, Z.; Geller, G.N.; Tsendbazar, N.-E.; See, L.; Griffiths, P.; Fritz, S.; Gong, P.; Herold, M.; Mora, B.; Obregón, A. Addressingthe Need for Improved Land Cover Map Products for Policy Support. Environ. Sci. Policy 2020, 112, 28–35. [CrossRef]

28. Nativi, S.; Santoro, M.; Giuliani, G.; Mazzetti, P. Towards a Knowledge Base to Support Global Change Policy Goals. Int. J. Digit.Earth 2019, 13, 188–216. [CrossRef]

29. Owers, C.J.; Lucas, R.M.; Clewley, D.; Planque, C.; Punalekar, S.; Tissott, B.; Chua, S.M.T.; Bunting, P.; Mueller, N.; Metternicht,G. Living Earth: Implementing National Standardised Land Cover Classification Systems for Earth Observation in Support ofSustainable Development. Big Earth Data 2021, 5, 368–390. [CrossRef]

30. Kavvada, A.; Metternicht, G.; Kerblat, F.; Mudau, N.; Haldorson, M.; Laldaparsad, S.; Friedl, L.; Held, A.; Chuvieco, E. TowardsDelivering on the Sustainable Development Goals Using Earth Observations. Remote Sens. Environ. 2020, 247, 111930. [CrossRef]

31. Whitcraft, A.K.; Becker-Reshef, I.; Justice, C.O.; Gifford, L.; Kavvada, A.; Jarvis, I. No Pixel Left behind: Toward Integrating EarthObservations for Agriculture into the United Nations Sustainable Development Goals Framework. Remote Sens. Environ. 2019,235, 111470. [CrossRef]

32. Giuliani, G.; Mazzetti, P.; Santoro, M.; Nativi, S.; Van Bemmelen, J.; Colangeli, G.; Lehmann, A. Knowledge Generation UsingSatellite Earth Observations to Support Sustainable Development Goals (SDG): A Use Case on Land Degradation. Int. J. Appl.Earth Obs. Geoinf. 2020, 88, 102068. [CrossRef]

33. Mander, Ü.; Müller, F.; Wrbka, T. Functional and Structural Landscape Indicators: Upscaling and Downscaling Problems. Ecol.Indic. 2005, 5, 267–272. [CrossRef]

34. Dennis, M.; Barlow, D.; Cavan, G.; Cook, P.; Gilchrist, A.; Handley, J.; James, P.; Thompson, J.; Tzoulas, K.; Wheater, C.P.; et al.Mapping Urban Green Infrastructure: A Novel Landscape-Based Approach to Incorporating Land Use and Land Cover in theMapping of Human-Dominated Systems. Land 2018, 7, 17. [CrossRef]

35. Whittaker, R.J.; Willis, K.J.; Field, R. Scale and Species Richness: Towards a General, Hierarchical Theory of Species Diversity.J. Biogeogr. 2001, 28, 453–470. [CrossRef]

36. Bontemps, S.; Defourny, P.; Radoux, J.; Van Bogaert, E.; Lamarche, C.; Achard, F.; Mayaux, P.; Boettcher, M.; Brockmann, C.;Kirches, G. Consistent Global Land Cover Maps for Climate Modelling Communities: Current Achievements of the ESA’s Land Cover CCI;European Space Agency: Frascati, Italy, 2013; pp. 9–13.

37. Haack, B.; Mahabir, R.; Kerkering, J. Remote Sensing-Derived National Land Cover Land Use Maps: A Comparison for Malawi.Geocarto Int. 2015, 30, 270–292. [CrossRef]

38. Randin, C.F.; Ashcroft, M.B.; Bolliger, J.; Cavender-Bares, J.; Coops, N.C.; Dullinger, S.; Dirnböck, T.; Eckert, S.; Ellis, E.; Fernández,N.; et al. Monitoring Biodiversity in the Anthropocene Using Remote Sensing in Species Distribution Models. Remote Sens.Environ. 2020, 239, 111626. [CrossRef]

39. Bojinski, S.; Verstraete, M.; Peterson, T.C.; Richter, C.; Simmons, A.; Zemp, M. The Concept of Essential Climate Variables inSupport of Climate Research, Applications, and Policy. Bull. Am. Meteorol. Soc. 2014, 95, 1431–1443. [CrossRef]

40. Pereira, H.M.; Ferrier, S.; Walters, M.; Geller, G.N.; Jongman, R.H.G.; Scholes, R.J.; Bruford, M.W.; Brummitt, N.; Butchart, S.H.M.;Cardoso, A.C.; et al. Essential Biodiversity Variables. Science 2013, 339, 277–278. [CrossRef]

41. Giuliani, G.; Egger, E.; Italiano, J.; Poussin, C.; Richard, J.-P.; Chatenoux, B. Essential Variables for Environmental Monitoring:What Are the Possible Contributions of Earth Observation Data Cubes? Data 2020, 5, 100. [CrossRef]

42. Lehmann, A.; Masò, J.; Nativi, S.; Giuliani, G. Towards Integrated Essential Variables for Sustainability. Int. J. Digit. Earth 2020, 13,158–165. [CrossRef]

43. Lucas, R.; Mitchell, A. Integrated Land Cover and Change Classifications. In The Roles of Remote Sensing in Nature Conservation;Springer: Cham, Switzerland, 2017; pp. 295–308, ISBN 978-3-319-64330-4.

Land 2022, 11, 615 19 of 21

44. Pettorelli, N.; Bühne, H.S.; Tulloch, A.; Dubois, G.; Macinnis-Ng, C.; Queirós, A.M.; Keith, D.A.; Wegmann, M.; Schrodt, F.;Stellmes, M.; et al. Satellite Remote Sensing of Ecosystem Functions: Opportunities, Challenges and Way Forward. Remote Sens.Ecol. Conserv. 2018, 4, 71–93. [CrossRef]

45. Moll, G.; Kay, K.; Maharjan, B. Remote Sensing & Classified Land Cover—Essential Land Use Decision Support Tools Using High-Resolution Imagery; Global Ecosystem Center: Washington, DC, USA, 2012.

46. Bateman, I.J.; Harwood, A.R.; Mace, G.M.; Watson, R.T.; Abson, D.J.; Andrews, B.; Binner, A.; Crowe, A.; Day, B.H.; Dugdale, S.;et al. Bringing Ecosystem Services into Economic Decision-Making: Land Use in the United Kingdom. Science 2013, 341, 45–50.[CrossRef] [PubMed]

47. Wulder, M.A.; Coops, N.C.; Roy, D.P.; White, J.C.; Hermosilla, T. Land Cover 2.0. Int. J. Remote Sens. 2018, 39, 4254–4284.[CrossRef]

48. Braun, D.; Damm, A.; Hein, L.; Petchey, O.L.; Schaepman, M.E. Spatio-Temporal Trends and Trade-Offs in Ecosystem Services:An Earth Observation Based Assessment for Switzerland between 2004 and 2014. Ecol. Indic. 2018, 89, 828–839. [CrossRef]

49. Price, B.; Kienast, F.; Seidl, I.; Ginzler, C.; Verburg, P.H.; Bolliger, J. Future Landscapes of Switzerland: Risk Areas for Urbanisationand Land Abandonment. Appl. Geogr. 2015, 57, 32–41. [CrossRef]

50. Verburg, P.H.; Alexander, P.; Evans, T.; Magliocca, N.R.; Malek, Z.; Rounsevell, M.D.; van Vliet, J. Beyond Land Cover Change:Towards a New Generation of Land Use Models. Curr. Opin. Environ. Sustain. 2019, 38, 77–85. [CrossRef]

51. Alloghani, M.; Al-Jumeily, D.; Mustafina, J.; Hussain, A.; Aljaaf, A.J. A Systematic Review on Supervised and UnsupervisedMachine Learning Algorithms for Data Science. In Supervised and Unsupervised Learning for Data Science; Berry, M.W., Mohamed,A., Yap, B.W., Eds.; Unsupervised and Semi-Supervised Learning; Springer International Publishing: Cham, Switzerland, 2020;pp. 3–21. ISBN 978-3-030-22475-2.

52. Costa, H.; Foody, G.M.; Boyd, D.S. Supervised Methods of Image Segmentation Accuracy Assessment in Land Cover Mapping.Remote Sens. Environ. 2018, 205, 338–351. [CrossRef]

53. Hong, D.; Yokoya, N.; Ge, N.; Chanussot, J.; Zhu, X.X. Learnable Manifold Alignment (LeMA): A Semi-Supervised Cross-Modality Learning Framework for Land Cover and Land Use Classification. ISPRS J. Photogramm. Remote Sens. 2019, 147, 193–205.[CrossRef]

54. Li, Y.; Tao, C.; Tan, Y.; Shang, K.; Tian, J. Unsupervised Multilayer Feature Learning for Satellite Image Scene Classification. IEEEGeosci. Remote Sens. Lett. 2016, 13, 157–161. [CrossRef]

55. Lary, D.J.; Zewdie, G.K.; Liu, X.; Wu, D.; Levetin, E.; Allee, R.J.; Malakar, N.; Walker, A.; Mussa, H.; Mannino, A.; et al. MachineLearning Applications for Earth Observation. In Earth Observation Open Science and Innovation; ISSI Scientific Report Series;Springer: Cham, Switzerland, 2018; pp. 165–218, ISBN 978-3-319-65632-8.

56. Talukdar, S.; Singha, P.; Mahato, S.; Shahfahad; Pal, S.; Liou, Y.-A.; Rahman, A. Land-Use Land-Cover Classification by MachineLearning Classifiers for Satellite Observations—A Review. Remote Sens. 2020, 12, 1135. [CrossRef]

57. Zhang, X.; Zhou, Y.; Luo, J. Deep Learning for Processing and Analysis of Remote Sensing Big Data: A Technical Review. BigEarth Data 2021, 1–34. [CrossRef]

58. Zhu, X.X.; Tuia, D.; Mou, L.; Xia, G.S.; Zhang, L.; Xu, F.; Fraundorfer, F. Deep Learning in Remote Sensing: A ComprehensiveReview and List of Resources. IEEE Geosci. Remote Sens. Mag. 2017, 5, 8–36. [CrossRef]

59. Chaves, M.E.D.; Picoli, M.C.A.; Sanches, I.D. Recent Applications of Landsat 8/OLI and Sentinel-2/MSI for Land Use and LandCover Mapping: A Systematic Review. Remote Sens. 2020, 12, 3062. [CrossRef]

60. Phiri, D.; Simwanda, M.; Salekin, S.; Nyirenda, V.R.; Murayama, Y.; Ranagalage, M. Sentinel-2 Data for Land Cover/Use Mapping:A Review. Remote Sens. 2020, 12, 2291. [CrossRef]

61. Wang, T.; Kazak, J.; Han, Q.; de Vries, B. A Framework for Path-Dependent Industrial Land Transition Analysis Using VectorData. Eur. Plan. Stud. 2019, 27, 1391–1412. [CrossRef]

62. Yoo, S.; Lee, J.; Farkoushi, M.G.; Lee, E.; Sohn, H.-G. Automatic Generation of Land Use Maps Using Aerial Orthoimages andBuilding Floor Data with a Conv-Depth Block (CDB) ResU-Net Architecture. Int. J. Appl. Earth Obs. Geoinf. 2022, 107, 102678.[CrossRef]

63. Potapov, P.; Hansen, M.C.; Kommareddy, I.; Kommareddy, A.; Turubanova, S.; Pickens, A.; Adusei, B.; Tyukavina, A.; Ying, Q.Landsat Analysis Ready Data for Global Land Cover and Land Cover Change Mapping. Remote Sens. 2020, 12, 426. [CrossRef]

64. Hoskins, A.J.; Bush, A.; Gilmore, J.; Harwood, T.; Hudson, L.N.; Ware, C.; Williams, K.J.; Ferrier, S. Downscaling Land-Use Datato Provide Global 30” Estimates of Five Land-Use Classes. Ecol. Evol. 2016, 6, 3040–3055. [CrossRef]

65. Wilby, R.L.; Wigley, T.M.L. Downscaling General Circulation Model Output: A Review of Methods and Limitations. Prog. Phys.Geogr. Earth Environ. 1997, 21, 530–548. [CrossRef]

66. Huth, R. Statistical Downscaling of Daily Temperature in Central Europe. J. Clim. 2002, 15, 1731–1742. [CrossRef]67. Coulibaly, P.; Dibike, Y.B.; Anctil, F. Downscaling Precipitation and Temperature with Temporal Neural Networks. J. Hydrometeorol.

2005, 6, 483–496. [CrossRef]68. Bogardi, I.; Matyasovzky, I. Estimating Daily Wind Speed under Climate Change. Sol. Energy 1996, 57, 239–248. [CrossRef]69. Huth, R. Downscaling of Humidity Variables: A Search for Suitable Predictors and Predictands. Int. J. Climatol. 2005, 25, 243–250.

[CrossRef]70. Müller-Wohlfeil, D.-I.; Bürger, G.; Lahmer, W. Response of a River Catchment to Climatic Change: Application of Expanded

Downscaling to Northern Germany. Clim. Chang. 2000, 47, 61–89. [CrossRef]

Land 2022, 11, 615 20 of 21

71. Wilby, R.L.; Hay, L.E.; Gutowski, W.J., Jr.; Arritt, R.W.; Takle, E.S.; Pan, Z.; Leavesley, G.H.; Clark, M.P. Hydrological Responses toDynamically and Statistically Downscaled Climate Model Output. Geophys. Res. Lett. 2000, 27, 1199–1202. [CrossRef]

72. Wood, A.W.; Leung, L.R.; Sridhar, V.; Lettenmaier, D.P. Hydrologic Implications of Dynamical and Statistical Approaches toDownscaling Climate Model Outputs. Clim. Chang. 2004, 62, 189–216. [CrossRef]

73. Lehmann, A.; Leathwick, J.R.; Overton, J.M. Assessing New Zealand Fern Diversity from Spatial Predictions of Species Assem-blages. Biodivers. Conserv. 2002, 11, 2217–2238. [CrossRef]

74. Dehn, M.; Bürger, G.; Buma, J.; Gasparetto, P. Impact of Climate Change on Slope Stability Using Expanded Downscaling. Eng.Geol. 2000, 55, 193–204. [CrossRef]

75. Le Page, Y.; West, T.O.; Link, R.; Patel, P. Downscaling Land Use and Land Cover from the Global Change Assessment Model forCoupling with Earth System Models. Geosci. Model Dev. 2016, 9, 3055–3069. [CrossRef]

76. Mancosu, E.; Gago-Silva, A.; Barbosa, A.; de Bono, A.; Ivanov, E.; Lehmann, A.; Fons, J. Future Land-Use Change Scenarios forthe Black Sea Catchment. Environ. Sci. Policy 2015, 46, 26–36. [CrossRef]

77. Bardossy, A.; Bogardi, I.; Matyasovszky, I. Fuzzy Rule-Based Downscaling of Precipitation. Theor. Appl. Climatol. 2005, 82, 119–129.[CrossRef]

78. Bürger, G.; Chen, Y. Regression-Based Downscaling of Spatial Variability for Hydrologic Applications. J. Hydrol. 2005, 311,299–317. [CrossRef]

79. Biau, G.; Zorita, E.; von Storch, H.; Wackernagel, H. Estimation of Precipitation by Kriging in the EOF Space of The Sea LevelPressure Field. J. Clim. 1999, 12, 1070–1085. [CrossRef]

80. Crow, W.T.; Wood, E.F.; Dubayah, R. Potential for Downscaling Soil Moisture Maps Derived from Spaceborne Imaging RadarData. J. Geophys. Res. Atmos. 2000, 105, 2203–2212. [CrossRef]

81. Lehmann, A.; Overton, J.M.; Leathwick, J.R. GRASP: Generalized Regression Analysis and Spatial Prediction. Ecol. Model. 2003,160, 165–183. [CrossRef]

82. Remm, K. Case-Based Predictions for Species and Habitat Mapping. Ecol. Model. 2004, 177, 259–281. [CrossRef]83. Aha, D.W. The Omnipresence of Case-Based Reasoning in Science and Application. Knowl. Based Syst. 1998, 11, 261–273.

[CrossRef]84. Karpatne, A.; Jiang, Z.; Vatsavai, R.R.; Shekhar, S.; Kumar, V. Monitoring Land-Cover Changes: A Machine-Learning Perspective.

IEEE Geosci. Remote Sens. Mag. 2016, 4, 8–21. [CrossRef]85. Ban, Y.F.; Gong, P.; Gini, C. Global Land Cover Mapping Using Earth Observation Satellite Data: Recent Progresses and Challenges.

Isprs J. Photogramm. Remote Sens. 2015, 103, 1–6. [CrossRef]86. Gellrich, M.; Zimmermann, N.E. Investigating the Regional-Scale Pattern of Agricultural Land Abandonment in the Swiss

Mountains: A Spatial Statistical Modelling Approach. Landsc. Urban Plan. 2007, 79, 65–76. [CrossRef]87. Nippel, T.; Klingl, T. Swiss Land Use in the European Context—Integration of Swiss Land Use Statistics with CORINE Land Cover; Swiss

Federal Statistical Office: Neuchâtel, Switzerland, 1998; p. 47.88. Conedera, M.; Tonini, M.; Oleggini, L.; Orozco, C.V.; Leuenberger, M.; Pezzatti, G.B. Geospatial Approach for Defining the

Wildland-Urban Interface in the Alpine Environment. Comput. Environ. Urban Syst. 2015, 52, 10–20. [CrossRef]89. Holloway, J.; Helmstedt, K.J.; Mengersen, K.; Schmidt, M. A Decision Tree Approach for Spatially Interpolating Missing Land

Cover Data and Classifying Satellite Images. Remote Sens. 2019, 11, 1796. [CrossRef]90. Ling, F.; Du, Y.; Li, X.; Li, W.; Xiao, F.; Zhang, Y. Interpolation-Based Super-Resolution Land Cover Mapping. Remote Sens. Lett.

2013, 4, 629–638. [CrossRef]91. Gardner, R.H.; Lookingbill, T.R.; Townsend, P.A.; Ferrari, J. A New Approach for Rescaling Land Cover Data. Landsc. Ecol. 2008,

23, 513–526. [CrossRef]92. Welcome to Python.Org. Available online: https://www.python.org/ (accessed on 22 February 2022).93. PyCharm: The Python IDE for Professional Developers by JetBrains. Available online: https://www.jetbrains.com/pycharm/

(accessed on 22 February 2022).94. Gdal. Available online: https://gdal.org (accessed on 22 February 2022).95. NumPy. Available online: https://numpy.org/ (accessed on 22 February 2022).96. Xlrd. Available online: https://xlrd.readthedocs.io/en/latest/ (accessed on 22 February 2022).97. Pandas—Python Data Analysis Library. Available online: https://pandas.pydata.org/ (accessed on 22 February 2022).98. High Performance Computing—EResearch—UNIGE. Available online: https://www.unige.ch/eresearch/en/services/hpc/

(accessed on 22 February 2022).99. Giuliani, G. Lulcdown. 2022. Available online: https://github.com/ggiuliani/LULCdown (accessed on 22 February 2022).100. Yareta—Portal. Available online: https://yareta.unige.ch/#/home/detail/6ab4b715-904f-4cb9-961c-6a25b4c1116b (accessed on

22 February 2022).101. Dendoncker, N.; Bogaert, P.; Rounsevell, M. A Statistical Method to Downscale Aggregated Land Use Data and Scenarios. J. Land

Use Sci. 2006, 1, 63–82. [CrossRef]102. Sherba, J.T.; Sleeter, B.M.; Davis, A.W.; Parker, O.; Sherba, J.T.; Sleeter, B.M.; Davis, A.W.; Parker, O. Downscaling Global

Land-Use/Land-Cover Projections for Use in Region-Level State-and-Transition Simulation Modeling. AIMS Environ. Sci. 2015, 2,623–647. [CrossRef]

Land 2022, 11, 615 21 of 21

103. Zheng, H.; Du, P.; Chen, J.; Xia, J.; Li, E.; Xu, Z.; Li, X.; Yokoya, N. Performance Evaluation of Downscaling Sentinel-2 Imagery forLand Use and Land Cover Classification by Spectral-Spatial Features. Remote Sens. 2017, 9, 1274. [CrossRef]

104. Grandini, M.; Bagli, E.; Visani, G. Metrics for Multi-Class Classification: An Overview. arXiv 2020, arXiv:2008.05756.105. West, T.O.; Page, Y.L.; Huang, M.; Wolf, J.; Thomson, A.M. Downscaling Global Land Cover Projections from an Integrated

Assessment Model for Use in Regional Analyses: Results and Evaluation for the US from 2005 to 2095. Environ. Res. Lett. 2014,9, 64004. [CrossRef]

106. Ivits, E.; Koch, B.; Blaschke, T.; Jochum, M.; Adler, P. Landscape Structure Assessment with Image Grey-values and Object-basedClassification at Three Spatial Resolutions. Int. J. Remote Sens. 2005, 26, 2975–2993. [CrossRef]

107. McGarigal, K.; Marks, B.J. FRAGSTATS: Spatial Pattern Analysis Program for Quantifying Landscape Structure; Gen. Tech. Rep.PNW-GTR-351; U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station: Portland, OR, USA, 1995;Volume 351, pp. 1–122. [CrossRef]

108. Uuemaa, E.; Roosaare, J.; Mander, Ü. Scale Dependence of Landscape Metrics and Their Indicatory Value for Nutrient andOrganic Matter Losses from Catchments. Ecol. Indic. 2005, 5, 350–369. [CrossRef]

109. Riitters, K.H. Downscaling Indicators of Forest Habitat Structure from National Assessments. Ecol. Indic. 2005, 5, 273–279.[CrossRef]

110. Levins, R. The strategy of model building in population biology. Am. Sci. 1966, 54, 421–431.111. Rockstrom, J.; Steffen, W.; Noone, K.; Persson, A.; Chapin, F.S.; Lambin, E.F.; Lenton, T.M.; Scheffer, M.; Folke, C.; Schellnhuber,

H.J.; et al. A Safe Operating Space for Humanity. Nature 2009, 461, 472–475. [CrossRef]112. Wulder, M.A.; Masek, J.G.; Cohen, W.B.; Loveland, T.R.; Woodcock, C.E. Opening the Archive: How Free Data Has Enabled the

Science and Monitoring Promise of Landsat. Remote Sens. Environ. 2012, 122, 2–10. [CrossRef]113. Zhu, Z.; Woodcock, C.E. Continuous Change Detection and Classification of Land Cover Using All Available Landsat Data.

Remote Sens. Environ. 2014, 144, 152–171. [CrossRef]114. Brown, D.G.; Walker, R.; Manson, S.; Seto, K. Modeling Land Use and Land Cover Change. In Land Change Science; Remote

Sensing and Digital Image Processing; Springer: Dordrecht, The Netherlands, 2012; pp. 395–409, ISBN 978-94-007-4306-9.115. Giuliani, G.; Dao, H.; De Bono, A.; Chatenoux, B.; Allenbach, K.; De Laborie, P.; Rodila, D.; Alexandris, N.; Peduzzi, P. Live

Monitoring of Earth Surface (LiMES): A Framework for Monitoring Environmental Changes from Earth Observations. RemoteSens. Environ. 2017, 202, 222–233. [CrossRef]

116. Purss, M.B.J.; Lewis, A.; Oliver, S.; Ip, A.; Sixsmith, J.; Evans, B.; Edberg, R.; Frankish, G.; Hurst, L.; Chan, T. Unlocking theAustralian landsat archive—From dark data to high performance data infrastructures. GeoResJ 2015, 6, 135–140. [CrossRef]

117. Ryan, B. The Benefits from Open Data Are Immense. Geospat. World 2016, 10, 72–73.118. Pasquarella, V.J.; Holden, C.E.; Kaufman, L.; Woodcock, C.E. From imagery to ecology: Leveraging Time series of all available

landsat observations to map and monitor ecosystem state and dynamics. Remote Sens. Ecol. Conserv. 2016, 2, 152–170. [CrossRef]119. Hermosilla, T.; Wulder, M.A.; White, J.C.; Coops, N.C.; Hobart, G.W.; Campbell, L.B. Mass Data Processing of Time Series Landsat

Imagery: Pixels to Data Products for Forest Monitoring. Int. J. Digit. Earth 2016, 9, 1035–1054. [CrossRef]120. Inglada, J.; Arias, M.; Vincent, A.; Tardy, B.D.M.; Michel, J. Large Scale Automatic Land Cover Map Production with Sentinel-2

Image Time Series: Current Status and Outlooks. 2016. Available online: https://www.jordiinglada.net/stok/LivingPlanet/LandCoverSlides.pdf (accessed on 22 February 2022).

121. Zioti, F.; Ferreira, K.R.; Queiroz, G.R.; Neves, A.K.; Carlos, F.M.; Souza, F.C.; Santos, L.A.; Simoes, R.E.O. A Platform for Land Useand Land Cover Data Integration and Trajectory Analysis. Int. J. Appl. Earth Obs. Geoinf. 2022, 106, 102655. [CrossRef]

122. Han, H.; Yang, C.; Song, J. Scenario Simulation and the Prediction of Land Use and Land Cover Change in Beijing, China.Sustainability 2015, 7, 4260–4279. [CrossRef]

123. Verburg, P.H.; Dearing, J.A.; Dyke, J.G.; van der Leeuw, S.; Seitzinger, S.; Steffen, W.; Syvitski, J. Methods and Approaches toModelling the Anthropocene. Glob. Environ. Chang.-Hum. Policy Dimens. 2016, 39, 328–340. [CrossRef]

124. Killough, B. CEOS Land Surface Imaging Analysis Ready Data (ARD) Description Document; CEOS: London, UK, 2016.125. Strobl, P.; Baumann, P.; Lewis, A.; Szantoi, Z.; Killough, B.; Purss, M.; Craglia, M.; Nativi, S.; Held, A.; Dhu, T. The Six Faces of the

Data Cube. In Proceedings of the 2017 Conference on Big Data from Space, Toulouse, France, 28 November 2017; pp. 32–35.126. Chatenoux, B.; Richard, J.-P.; Small, D.; Roeoesli, C.; Wingate, V.; Poussin, C.; Rodila, D.; Peduzzi, P.; Steinmeier, C.; Ginzler,

C.; et al. The Swiss Data Cube, Analysis Ready Data Archive Using Earth Observations of Switzerland. Sci. Data 2021, 8, 295.[CrossRef]

127. Giuliani, G.; Chatenoux, B.; Bono, A.D.; Rodila, D.; Richard, J.-P.; Allenbach, K.; Dao, H.; Peduzzi, P. Building an EarthObservations Data Cube: Lessons Learned from the Swiss Data Cube (SDC) on Generating Analysis Ready Data (ARD). Big EarthData 2017, 1, 100–117. [CrossRef]

128. Gómez, C.; White, J.C.; Wulder, M.A. Optical Remotely Sensed Time Series Data for Land Cover Classification: A Review. ISPRSJ. Photogramm. Remote Sens. 2016, 116, 55–72. [CrossRef]